>make ext4 filesystem (or any extX)
>by default it uses 5% "reserved space" for root and no defragmentation
>on a 2 TB drive that's like 100 GB unusable storage
Why would you want this? I just use mkfs.ext4 -m 0 (or tune2fs for existing) because I really want the storage and not this stupid overhead. With 0% the filesystem creation only uses <100 MB.
>>55307896
How stable and fast is ext4 compared to btrfs?
>>55307896
Doesnt the "no fragmentation" meme technically come with performance losses which are just as comparable over time?
>>55308008
Explain please.
>>55307940
I thought btrfs was unstable. ext4 is pretty solid.
>>55307977
No fragmentation means less chance of mechanical failure on a hard drive due to the files being in one place.
>>55308047
But if it basically takes the same amount of mechanical work to ensure contiguous files, whats the difference in load? Can you show me any reports, studies, abstracts on it?
>>55308310
Not really, there might be some structure which contains a list of free memory (maybe this uses some of the reserved 5%), and this structure is probably efficient in terms of adding and removing entries (i.e. you can find contiguous memory of a given size in O (log(n)) time and perhaps free memory in constant time). Defragmentation is almost always O(n) - even if an implementation is faster than this, its speed will be significantly reduced because of the fact that the contents have to be copied
>>55307896
The reserved space is so that processes running as non-root users (aka pretty much all of them) can't actually fill up the last free bytes on disk and prevent root from doing things to fix the system.
>>55307896
it should be 2 MiB anon
